A fair warning before we start. This article has an expiration date. Not the principle. The principle is permanent: marketing copy should sound like someone with conviction wrote it. But the specific tells I catalog here reflect LLM-generated text I’ve seen from 2024 to 2026. AI writing patterns will evolve. Signals will shift. A year from now, some of these tells will be less common and new ones will replace them. Treat the examples as a snapshot.

The Problem Isn’t AI. It’s Default AI.

One final preface — I use AI tools every day. This isn’t an anti-AI paper. It’s an anti-laziness paper.

AI-generated copy isn’t bad because AI is bad. It’s bad because most marketing content people accept the first draft. The default output of any LLM typically optimizes for completeness and inoffensiveness. Not for voice. Not for conviction. Not for story. The result is copy that says everything and means nothing. Technically correct. Emotionally flat. The kind of writing that fills a page without ever making a point.

The real cost is harder to measure but easy to feel. Audiences are developing a sixth sense for this stuff. A 2024 study by the Nuremberg Institute for Market Decisions found that when consumers were told content was AI-generated, they rated it as less credible, less appealing, and less natural — even when the content was identical to what another group was told was human-made. The researchers described it as a “trust penalty.” People don’t need to articulate why something feels off. They just know. And when they know, trust drops.

The Current Tells

George Orwell wrote in 1946 that bad writing spreads by imitation. He was talking about political language, but he could have been describing every B2B SaaS landing page written since ChatGPT launched. His advice was to ask yourself, before every sentence: What am I trying to say? What words will express it? Could I put it more shortly? Have I said anything that is avoidably ugly?

LLMs skip all four questions. That’s why the output has recognizable patterns.

Emdashes as connective tissue. AI loves emdashes. They connect clauses without committing to a sentence structure. In marketing copy, they’re almost always a sign that a sentence should have been two sentences — or that the writer let the model meander instead of deciding what the sentence was actually about.

Comma chains. Long sentences with three or four clauses joined by commas. Technically grammatical. Practically unreadable. If a sentence has more than one comma, it probably needs to be split.

The buzzword layer. Words that sound impressive but communicate nothing: seamless, robust, leverage, harness, empower, unlock, streamline, transformative, holistic, cutting-edge, next-generation, best-in-class. Every competitor in every category uses these words. That’s exactly how you know they’re empty. If a word works equally well in your copy and your competitor’s copy, it isn’t doing anything.

Circular structure. Paragraph opens with a claim. Middle restates it with different words. End restates it again as a conclusion. Three sentences, one idea, zero progression. The paragraph feels complete but never actually went anywhere.

The hedging opener. “It’s worth noting that…” / “It’s important to remember…” / “Ultimately…” These add words without adding meaning. Cut them and the sentence gets better every time.

Symmetrical structure. AI loves parallel construction to a fault. Three pillars, each with exactly three bullets, each bullet the same length. Real ideas aren’t that tidy. I’ll admit — I love the rule of three myself. Three bullets. Three pillars. My PM teases me about it constantly. But just because you see the structure doesn’t mean it’s AI. In aggregate with other factors, though, it’s a dead giveaway.

Before and After

Here’s the circular structure problem, pulled from a draft I rewrote last year.

AI default:

Digital transformation requires organizations to rethink how they approach process automation. By reimagining their automation strategy, companies can unlock new efficiencies and drive meaningful change. This transformation of processes through automation enables organizations to achieve their digital objectives more effectively.

Three sentences, one idea, stated three different ways. Here’s the rewrite:

Rewritten:

Most automation projects start with the wrong question. Teams ask “what can we automate?” instead of “what’s actually broken?” Start with the process that’s costing you the most time, money, or frustration. Automate that one first.

Same topic. Progression from problem to recommendation. The rewrite has something the original doesn’t: a position.

How to Use AI Effectively

Here is how I tackle the synergy between the marketer and AI. I put rules in place — rules I either manually enforce when editing, or give AI as guardrails before writing or during revision.

  • Short sentences. If a sentence has more than one comma, it probably needs to be two sentences. Read it out loud. If you run out of breath, you’ve made the case for me.
  • Second person for customer-facing copy. “You” and “your” force specificity. “Organizations can leverage” is vague. “You can connect your ERP data without replication” is concrete.
  • Problem before solution. Don’t describe the product until the reader knows why they should care. Lead with the pain.
  • Bullets: 3 is ideal, 5 is the ceiling. If you have 6 items, cut one or split into two groups.
  • Positive framing over negative. Say what you do, not what the competitor doesn’t. “K2 runs where your data lives” beats “Unlike cloud-only platforms that restrict your deployment options.”
  • Read it out loud. If you wouldn’t say it in a meeting, don’t write it in copy.
  • One idea per paragraph. If a paragraph makes two points, it’s two paragraphs.

The Underlying Principle

Good copy has a point of view. It takes a position. It says “this matters and here’s why” instead of “here are some things to consider.” Voice is the one thing AI can’t fake on the first pass.

Mailchimp understood this before most of the industry caught on. Their content style guide codifies a voice that is informal, clear, and human. One of their core principles: “We are plainspoken. We understand the world our customers are living in: one muddled by hyperbolic language, upsells, and over-promises. We strip all that away and value clarity above all.”

37signals takes it further. DHH and Jason Fried write under their own names. Their marketing reads like opinion pieces, not product pages. As DHH put it: “I substitute my name for whatever the company says, and I feel like it’s my credibility on the line.” That accountability shapes every word.

The identity of a brand comes from the specific choices you make about what to say and what not to say. Default AI can’t make those choices. It includes everything. Inclusion is its whole design. Voice comes from exclusion. From deciding what doesn’t belong.

Work With AI Without Losing Your Voice

AI is a drafting tool, not a finishing tool. Use it to get ideas on paper. Then rewrite.

The value of AI for product marketers isn’t in the output. It’s in the speed of the first draft. The difference between staring at a blank page for 30 minutes and having a rough draft you can tear apart in 10. You still have to do the last mile. That last mile is where voice lives.

Build a style guide that AI can follow. Codify your banned words. Write down your structural rules. Document your voice preferences. My banned list includes em dashes, “seamless,” “robust,” “harness,” “empower,” and about 20 others. Yours will be different. The act of building the list forces you to decide what your voice isn’t.

The best AI-assisted copy is invisible. Nobody should be able to tell a machine helped write it. If someone reads your landing page and thinks “this sounds like ChatGPT,” you stopped too early. The draft was a starting point. You treated it as a finish line.

Sources referenced: George Orwell, “Politics and the English Language” (1946) · Mailchimp Content Style Guide · Nuremberg Institute for Market Decisions, “Transparency Without Trust” (2024) · Ann Handley, Everybody Writes, 2nd ed. (2022) · 37signals, “Founder-led Marketing”